Date of Award

Winter 2020

Project Type


Program or Major

Civil Engineering

Degree Name

Master of Science

First Advisor

Thomas P Ballestero

Second Advisor

James P Malley

Third Advisor

James J Houle


This research project analyzes the hydrology of two Green Stormwater Infrastructure (GSI) systems located at the University of New Hampshire (UNH) campus in Durham, NH, and compares field data to modeling results of a calibrated Storm Water Management Model (SWMM) model of each system. The studied systems were a Philadelphia Tree Trench and an Infiltration Trench, located in Parking Lot A and Parking Lot E, respectively. The Stormwater Center at UNH monitored the system wells, precipitation, and collected data since system constructions, to analyze the infiltration behavior.The fundamental reason for this research is that the SWMM model only computes infiltration out the bottom of GSI systems whereas field data indicate that significant additional water infiltrates horizontally out the system walls. The objective of this research is to understand how well the model results match the observed system performance. The methods used in this evaluation were the visual comparison of observed water volume versus model water volume; the Mean Square Error (RMSE), and the Nash-Sutcliffe equation (NSE). The model was originally planned to be calibrated by changing only infiltration parameters in the system, according to the Green-Ampt method of infiltration. A sensitivity analysis showed that the hydraulic conductivity was the most relevant parameter in the seepage loss calculation in SWMM. However, changing model infiltration area to include sidewalls in both systems significantly improved the results. This was found to be necessary due to SWMM not considering horizontal infiltration for the seepage loss calculations. The hydraulic conductivity values of the calibrated model were below the expected values for the soil types present in the field, even with the correction of the infiltration area. This calibration concluded that SWMM predicts infiltration rates 33% of the rates expected for the soil types on average, but very similar infiltration rates when compared to the ones measured on the field for these systems. SWMM predicted modeled infiltrated volumes 14% of observed volumes when using storage units to model infiltration systems. Final NSE and RMSE values were improved in the calibration, but not as expected for goodness-of-fit. Two methods were tested in the attempt to obtain modeled infiltrated volumes matching the ones observed in the field. The first one was to model the system as a LID control option. It was concluded to be ineffective when modeling the systems in this study, as this method underpredicted infiltrated volumes for some storms events (around 59%), and overpredicted for others (around 149%). This may be due to the proportion of runoff volume entering the system in the model not matching the one observed in reality when using LID control options to model infiltration systems. The last method was to calibrate the model with the addition of a fictitious underdrain to help improve infiltration in the systems. This was concluded to be the best option, as the modeled infiltrated volumes matched almost 100% the ones observed for both systems. This method presented a significant improvement in final NSE and RMSE values when compared to the original calibration process. The water flowing through the fictitious underdrain would simulate the water flowing through the sidewalls of the system in reality. Therefore, the modeled volume of water flowing out of the system through the fictitious underdrain would simulate the observed infiltrated volume of water flowing through the sidewalls of the system in reality. However, this is not a feasible method to implement, as it is not practical to estimate the diameter of the fictitious underdrain during the design phase of new systems. The conclusion of this study is that the calibration was only possible due to the availability of observed data. When comparing modeled results to observed data, it was noticed that it is important to consider parameters other than infiltration rate when modeling GSI systems in SWMM. This means that SWMM models of GSI systems are incapable of adequately representing lateral infiltration, when considering only the available infiltration parameters in SWMM.